The karst area in South China is notable for its fragile vegetation and unique flora with many narrow endemics, and is one of the most threatened biodiversity hotspots in the world. However, the biodiversity of acid soil areas scattered in the karst area have previously been underestimated. With a recent increase of new discoveries of plants and animals, the conservation of biodiversity in these acid soil areas has become urgent. This study deals with a new species, Impatiens longlinensis. The new species is similar to I. yui S.H.Huang and I. lasiophyton Hook.f. in having pubescence, two lateral sepals, funnelform lower sepal, and linear capsule. But it can be distinguished by having orbicular to obovate dorsal sepals and oblong to elliptic lower and upper lobes of lateral united petals. Moreover, molecular data and micro-morphological evidence also support that the species is new to science. The new species is only known from a valley near the summit of Mt. Jinzhongshan within an area of less than 5 km². Based on the Red List Categories and Criteria developed by the International Union for Conservation of Nature, and the threat posed by exploitation for tourism development, we categorize it as Critically Endangered (CR).

Taxonomic Sampling— to confirm the phylogenetic placement of the new species, 153 species of Impatiens were sampled. Hydrocera triflora (L.) Wight & Arn. (Balsaminaceae), Marcgravia umbellata L. and Norantea guianensis Aubl. (Marcgraviaceae) were included as outgroups (Yuan et al. 2004; Janssen et al. 2006; Yu et al. 2016). Sequences were downloaded from GenBank, except for I. longlinensis and I. yui, that were newly generated in this study. GenBank accession numbers are listed in Table S1.

DNA Extraction, PCR Reaction and Sequencing—Three molecular markers were used including ITS (ITS-1, 5.8S, and ITS-2), atpB-rbcL, and trnL-F region (trnL intron, and trnL [UAA] 3’ exon-trnF [GAA] intergenic spacer). Total genomic DNA was extracted from silica gel-dried leaves using a modified CTAB protocol (Doyle and Doyle 1987). Primers and PCR protocols for ITS, atpB-rbcL and trnL-F were derived from White et al. (1990), Janssens et al. (2006) and Taberlet et al. (1991), respectively. The PCR products were purified using a GFX^TMPCR DNA and gel band purification kit (Amersham Pharmacia Biotech, Piscataway, New Jersey). Sequencing reactions were carried out using an ABI Prism Bigdye terminator cycle sequencing kit (Applied Biosystems, Foster City, California). Products were analyzed on an ABI3730xl automated DNA sequencer (Beckman Coulter Genomics, UK).

Phylogenetic Analysis—Sequences were aligned using the default parameters in Clustal X v.1.83 (Thompson et al. 1997) and were further adjusted manually in BioEdit v.7.0 (Hall 1999). Four difficult-to-align regions in trnL-F (encompassing 73 sites) and one difficult-to-align region in atpB-rbcL (encompassing 42 sites) were excluded from the analysis. Maximum parsimony (MP) and Bayesian inference (BI) were used to analyze the ITS and plastid data sets. The MP analyses were carried out in PAUP* v.4.0b10 (Swofford 2003). Heuristic searches were conducted with 1000 replicates of random addition, one tree held at each step during stepwise addition, tree-bisection-reconnection (TBR) branch swapping, MulTrees in effect, and steepest descent off. Bootstrapping was conducted with 1000 replicates with 10 random taxon additions and heuristic research options. The BI analyses were carried out in MrBayes v.3.0b4 (Ronquist and Huelsenbeck 2003). Each of the three regions (ITS, atpB-rbcL, and trnL-F) was assigned to its own nucleotide substitution model (GTR+I+G model for all), as determined by the Akaike information criterion (AIC) in Modeltest v.3.06 (Posada and Crandall 1998).